Prognostic and Risk Stratification Value of Lesion MACC1 Expression in Colorectal Cancer Patients

The up-regulated metastasis-associated in colon cancer 1 (MACC1) expression and its clinical significance has been explored in a varity of malignancies. In this study, lesion MACC1 expression in 503 CRC patients (Ncolon = 332, Nrectal = 171) were analyzed with immunohistochemistry, and its correlation with clinical parameters, patient survival, and its impact on prognostic stratification were evaluated. Data revealed the survival of patient with MACC1high is markedly worse than that of MACC1low (mean overall survival: 80.1 vs. 90.4 months; p = 0.001) and is an independent prognostic predictor (hazard ratio = 1.533; p = 0.005). More importantly, for the first time, we demonstrated that MACC1 status exhibited a significantly prognostic power for stratified clinical parameters such as patient age and gender, particular TNM status, and distinct AJCC disease stage. In summary, our findings indicated that MACC1 is a valuable prognostic and risk stratification biomarker for colorectal cancer patients.


INTRODUCTION
Cancer is one of the major public health problems and the leading cause of death worldwide. In this scenario, the incidence and mortality of the colorectal cancer (CRC) has been observed on an upward trend during the last decade in China. There were 376,300 new CRC cases and 191,000 deaths estimated for the year of 2015, making CRC among the most commonly diagnosed and cancer-related death in our country (1,2).
Cancers have developed comprehensive strategies such as aberrant induction of cancer-promoting molecule expression to counteract host anti-tumor responses for malignant cell survival and metastasis, and finally result in disease progression (3). Among tremendous kinds of tumor promoting antigens, metastasis-associated in colon cancer 1 (MACC1) has been found to play pivotal roles in cancer tumorigenesis and metastasis, and its relevance in chemoresistance has also been reported (4)(5)(6)(7).
In CRC patients, it has been found that tumor lesion MACC1 expression significantly increased in relation to its non-tumorous adjacent tissues. Enhanced MACC1 expression has been reported to be significantly related to tumor metastasis and worse disease outcome, and to be an early risk factor for cancer patients (8,9). Moreover, higher circulating MACC1 transcripts and soluble MACC1 proteins have also been found relating to unfavorable prognosis for cancer patients (10,11). Other than CRC, as it has been investigated, the prognostic value of MACC1 has been further proved in other malignancies such as hepatocellular cancer (12), ovarian cancer and breast cancer (13,14). In this context, lesion MACC1 expression has been suggested as a poor prognostic biomarker in patients with nonsmall cell lung cancer (15). Tan et al. (11) reported that serum MACC1 levels can not only discriminate breast cancer patients from normal controls, but can also have high MACC1 levels that is related to patient worse disease-free survival. Burock et al. (10) presented that peripheral circulating MACC1 transcripts is a valuable diagnostic and prognostic factor for gastric cancer patients.
However, being the heterogeneity of cancers, the disease outcome can vary markedly even among patients with the same tumor-lymph nodes-metastasis (TNM) status or The American Joint Committee on Cancer (AJCC) stages (16). Therefore, in order to improve disease outcome prediction in distinct subpopulations of cancer patients, the importance of risk stratification with other prognostic factors has to be recognized (17,18). In this study, we assessed the impact of lesion MACC1 expression on survival and prognostic stratification value in a large cohort of 503 CRC patients.  (19). Follow-up data were available for 499 patients until the last follow-up on December 2016. The median follow-up was 53.0 months (range: 3∼137 months), and 180 deaths of the CRC patients occurred during the period. Patient overall survival was calculated from the date of surgical operation  Table 1.

Immunohistochemistry (IHC) and Evaluation of Staining
Immunohistochemistry were performed with 4 µm sections of paraffin-embedded tissues on polylysine coated slides. Slides were deparaffinized with xylene and rehydrated through gradient ethanol. Antigen retrieval was performed with 10 mM sodium citrate buffer (pH 6.0), and endogenous peroxidase activity was blocked with 3% H 2 O 2 . Then incubated with anti-MACC1 mAb (CL0856) (1:500, Thermo Fisher, Rockford, IL, USA) overnight at 4 • C. After they were thoroughly washed with 0.01 M phosphatebuffered saline (PBS), incubated with Envision anti-mouse and visualized by DAB development with a Dako EnVison kit (Dako, Glostrup, Denmark). Finally, all slides were counterstained with hematoxylin and mounted with glycerol gelatin. MACC1 expression index was assessed according to the intensity of staining and percentage of the MACC1 positive tumor cells. The staining intensity and proportion of MACC1 expression were scored independently by two observers who did not have access to the patients' information, and an average score of all samples were obtained. The staining intensity score was defined as: 0 (no staining); 1 (weak to moderate staining), and 2 (strong staining). MACC1 index of each slide was determined by multiplying the score of staining intensity and the percentage of MACC1 expression. The tumor MACC1 expression index in this study ranges 0∼2.
The representatives of immunohistochemistry are shown in Figure 1A and the distribution of the index of MACC1 expression is shown in Figure 1B. The range of the index was from 0 to 1.98 (median = 0.950). In addition to the median level, an appropriate cut-point determined by the receiver operating characteristic (ROC) curve to stratify the MACC1 high and MACC1 low groups was performed as recommended by Rohr et al. (20). In this study, the optimum cut-off for strata of MACC1 expression was determined by the ROC curve (between the CRC patients who survived and died) with the maximum of sensitivity and specificity according to the Youden's Index, and an optimum cut-off = 1.04 was obtained for the cohort (Figure 1B).
Therefore, in this study, two thresholds of MACC1 expression (median and cut-off) were analyzed for the CRC patient survival. Patients with MACC1 index> (median = 0.950) were defined as MACC1 high and ≤0.950 as MACC1 low . Moreover, we further analyzed the patients whose MACC1index >ROC based (cutoff = 1.04) were defined as MACC1 high and ≤1.04 as MACC1 low .

Statistical Analysis
Statistical analysis was performed with the SPSS 13.0 statistical software package (SPSS, Inc., Chicago, IL, USA). ROC curve was performed and the cut-off value was determined by Youden's index. The relationship between MACC1 status and CRC patient clinicopathological parameters were performed with Fisher's exact test or Chi-square test. Survival curves were performed with the Kaplan-Meier method, and differences between survival were compared with the log-rank test. The significance of variables for survival was conducted with the Cox proportional hazards model in multivariate analysis. P < 0.05 (two-tailed) was considered statistically significant.

Relationship Between MACC1 Levels and Clinical Variables in CRC Patients
According to the median level of MACC1 index (median = 0.950), among 503 CRC patients, there were 221 MACC1 high and 282 MACC1 low CRC patients, respectively. Based on the optimum cut-off by the Youden's Index, there were 98 MACC1 high and 405 MACC1 low CRC patients in this study.
The relationship between MACC1 expression and clinical variables of CRC patients is detailed in Table 1. The data revealed that patients with MACC1 high were more frequently observed in colon cancer patients than those in rectal cancer patients with both thresholds (p median = 0.037 and p cut−off = 0.017). The Frontiers in Oncology | www.frontiersin.org higher percentages of MACC1high were also observed in the CRC patients that had died (p median = 0.004 and p cut-off < 0.001). Among the whole cohort of CRC patients, an increasing trend of MACC1 high were found among CRC patients with advanced pT, pN, pM categories and advanced AJCC disease stages ( Table 1).

MACC1 Levels Related to Survival in CRC Patients
We, then evaluated the clinical significance of MACC1 levels to survival in CRC patients. Data showed that CRC patients with MACC1 high (>median) have a significantly worse overall mean of survival (OS) than those with MACC1 low (80.1 vs. 90.4 months), and the 5-year survival rate (SR) for the two groups is 58.9 vs. 70.2% (p = 0.001; Figure 2A). Moreover, the overall mean of worse survival and 5-year SR were also observed for the patients with MACC1 high expression in either colon cancer (n = 328; mean OS: 84.0 vs. 92.1 months; p = 0.007; Figure 2B) or rectal cancers (n = 171; mean OS: 53.0 vs. 79.5 months; p = 0.019; Figure 2C).
In addition, we also analyzed the effects of the MACC1 index with cut-off on the CRC patient survival. Data revealed that worse survival had been observed for CRC patients with MACC1 above the cut-off than those with MACC1 below the cut-off (67.3 vs. 92.0 months; p < 0.001), and the 5year SR for the two groups was 47.5 vs. 69.7% (p < 0.001; Supplementary Figure 1A). Similarly, worse survival and 5year SR were also observed for the patients with MACC1 above the cut-off in either colon cancer (71.9 vs. 94.4 months; p < 0.001; Supplementary Figure 1B) or rectal cancers (43.8 vs. 75.0 months; p < 0.001; Supplementary Figure 1C).
To mitigate the biological and clinical heterogeneity of samples and patients, we further assessed the value of MACC1 index (median) for survival among patients with a particular AJCC stage I, II, III, and IV, respectively. Status of MACC1 high and MACC1 low can separate the Kaplan-Meier curves for    Figure 2D). Next, we analyzed the prognostic value of MACC1 expression and other clinical parameters with the Cox's proportional hazards model. Data revealed that, in addition to subtypes of cancer and patient age, MACC1 index (media n = 0.950) is an independent prognostic factor for CRC patients (HR = 1.533, p = 0.005; Table 2). When the levels of MACC1 expression was grouped as above or below the cut-off levels, MACC1 expression status can also be an independent prognostic factor for CRC patient prognosis prediction (HR = 2.024, p < 0.001; Supplementary Table 1).

Prognostic Stratification Effects of MACC1 Levels in CRC Patients
Moreover, we analyzed the prognostic stratification effects of MACC1 index (median) on various clinical parameters including sub-histological tumor type, gender of the patient and age, TNM status and AJCC stages. The data demonstrated that MACC1 high and MACC1 low is of great power in survival when these variables were stratified. As shown in Table 3, MACC1 high and MACC1 low can further notably separate the survival curve among patients with colon (p < 0.001) or rectal cancer (p = 0.017), male (p = 0.006) or female (p = 0.033), elder (p = 0.009) or younger (p = 0.046), T 1+2 (p = 0.038) or T 3+4 (p = 0.014), N 0 (p = 0.012) or N 1+2 (p = 0.048), M 0 (p = 0.002), AJCC I+II (p = 0.022), or AJCC III+IV (p = 0.041). To be noted, the stratification survival analysis does not reach a statistical significance for patients with status of M 1 (p = 0.549), which may be due to the limited size of only 16 patients. However, an indisputable trend was observed for the MACC1 high toward a worse outcome to stratified clinical parameters. Similar findings were obtained with the threshold value of MACC1 index with the cut-off (Supplementary Table 2).

DISCUSSION
MACC1 gene was first identified by Stein et al. (4) in 2009 in patients with colon cancer. Thereafter, a wealth of studies have been carried out in variety of malignancies, strengthening the potential application of both MACC1 transcripts and protein expression as a novel prognostic indicator, and MACC1 as a therapeutic target was recommended for cancers (21). The relevance of MACC1 expression including its genetic and proteomic has been explored in a large body of studies. In this context, higher levels of circulating MACC1 mRNA, peripheral soluble or tumor lesion protein expression was significantly associated with poor survival in patients such as lung cancer (22,23), gastric cancer (24), glioma (25), cervical cancer (26), hepatocellular, and renal pelvis carcinoma (12,27).
Mechanisms of MACC1 in cancer development and progression have been reported in in vitro cell model and pre-clinical murine models. MACC1 has been found to enhance gastric tumor cell migration, invasion and epithelialmesenchymal transition (EMT) in vitro (28). Authors further addressed that MACC1 overexpression favors tumor growth and promotes tumor metastasis in an athymic mice model. Multiple signal pathways such as various microRNA (29,30), lncRNA (31,32), circular RNA (33), and molecules such as deleted in breast cancer 1 (34), statin and rottlerin (35), have been observed in the regulation of MACC1 expression. Based on these findings, MACC1 as a potential valuable therapeutic target has been proposed.
In the current study, we evaluated the prognostic significance of tumor lesion MACC1 expression in 503 CRC patients. Our data revealed that MACC1 high is strongly associated with poor disease outcome and can be an independent prognostic factor. More importantly, MACC1 status could further improve the prognostic power for stratified clinical parameters, such as basic patient characteristics (patient age and gender) or distinct pathological factors (cancer subtype, TNM status and AJCC stages). Our data clearly demonstrated that, patients with MACC1 high have a significantly shorter survival rate than those with MACC1 low in various stratified parameters. These parameters include patient age, gender, colon or rectal cancer, TNM status and AJCC stage. To be noted, our results showed that threshold with the cut-off value based on ROC was much powerful in association with CRC patient survival than that of the median of the MACC1 index. As Rohr et al. (20) suggested in their recent study that the cut-off determination by ROC is an important step to ensure the future use of MACC1 protein expression, and that this can be more easily adapted to clinical practice. However, this may cause bias due to the ROC analysis and the prognostic power of the MACC1 expression in CRC patients were performed on the same cohort.
Moreover, our data revealed that the percentage of MACC1 high was more frequently observed in colon cancer patients. Patients with MACC1 high expression in both colon cancer and rectal cancer showed a significantly worse prognosis than those with MACC1 low . In line with this, similar findings have been obtained by Zhu et al. (36) indicate that the overexpression of MACC1 was associated with poor survival in patients with colonic adenocarcinoma, and that MACC1 status can be an independent predictor of prognosis in patients with colonic adenocarcinoma. Moreover, a study by Rohr et al. (20) recently presented that status of MACC1 expression could stratify stage II colon cancer patients with unfavorable proficient mismatch repair (pMMR) status, and a distinct stage II colon cancer patients with pMMR/MACC1 low had a favorable prognosis similar to those with deficient mismatch repair (dMMR). In breast cancer, when estrogen receptor (ER) status were stratified, MACC1 was of prognostic value for both ER-negative and ER-positive patients (37). However, more investigation is needed to solidify the significance of the prognostic stratification of MACC1 expression in cancers.
In conclusion, we provided the evidence that tumor lesion MACC1 status is a clinical prognostic biomarker for patients with CRC and that it is also an improved prognostic significance for distinct stratified clinical parameters.